Load and Explore Cifar10 Dataset

Image Classification Task

Cifar10 is a famous computer-vision dataset used for object recognition.

The dataset consists of:

• 32x32 pixel colored images

• 10 classes

• 6,000 images per classes

• 50,000 images in the training set

• 10,000 images in the test set

Imports

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import tensorflow as tf
import matplotlib
import matplotlib.pyplot as plt
import numpy as np

Download and Load Cifar10 Dataset

(x_train, y_train), (x_test, y_test) = tf.contrib.keras.datasets.cifar10.load_data()

Training Tensor Shape

x_train.shape

(50000, 32, 32, 3)

Testing Tensor Shape

x_test.shape

(10000, 32, 32, 3)

Ploting Helper Function

def plot_10_by_10_images(images):

    # figure size
    fig = plt.figure(figsize=(10,10))

    # plot image grid
    for x in range(10):
        for y in range(10):
            ax = fig.add_subplot(10, 10, 10*y+x+1)
            plt.imshow(images[10*y+x])
            plt.xticks(np.array([]))
            plt.yticks(np.array([]))
    plt.show()

Explore Cifar10 Dataset

plot_10_by_10_images(x_train[:100])

Next Lesson

SqueezeNet Architecture

• AlexNet-level accuracy with 50x fewer parameters

• CNN Squeeze layers and delayed downsampling